LGAICLSep 15, 2023

CoCA: Fusing Position Embedding with Collinear Constrained Attention in Transformers for Long Context Window Extending

arXiv:2309.08646v330 citationsh-index: 14Has Code
Originality Highly original
AI Analysis

This addresses the challenge of scaling context windows in large language models, which is crucial for applications requiring long-range dependencies, though it appears incremental as it builds on existing transformer architectures.

The paper tackles the problem of long context window extrapolation in transformers by proposing CoCA, a novel attention mechanism that integrates Rotary Position Embedding with self-attention through a collinear constraint, enabling a GPT model trained on 512-length context to extend up to 32K without fine-tuning.

Self-attention and position embedding are two key modules in transformer-based Large Language Models (LLMs). However, the potential relationship between them is far from well studied, especially for long context window extending. In fact, anomalous behaviors harming long context extrapolation exist between Rotary Position Embedding (RoPE) and vanilla self-attention unveiled by our work. To address this issue, we propose a novel attention mechanism, CoCA (Collinear Constrained Attention). Specifically, we enforce a collinear constraint between $Q$ and $K$ to seamlessly integrate RoPE and self-attention. While only adding minimal computational and spatial complexity, this integration significantly enhances long context window extrapolation ability. We provide an optimized implementation, making it a drop-in replacement for any existing transformer-based models. Extensive experiments show that CoCA performs extraordinarily well in extending context windows. A CoCA-based GPT model, trained with a context length of 512, can seamlessly extend the context window up to 32K (60$\times$), without any fine-tuning. Additionally, by dropping CoCA in LLaMA-7B, we achieve extrapolation up to 32K within only 2K training length. Our code is publicly available at: https://github.com/codefuse-ai/Collinear-Constrained-Attention

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes